Comparing Plasma Donor-Derived Cell-free DNA to Indication Kidney Biopsy Tissue Gene Expression: Toward Understanding the Molecular Equivalents of Non-Invasive Tests

Halloran and colleagues describe the results of plasma donor-derived cell-free DNA (dd-cfDNA) measurements from the prospective Trifecta study that examines the relationship between dd-cfDNA measured at the time of indication biopsy, and its relationship to global transcriptomic profiling from kidney transplant biopsies assessed using microarrays.¹ Their findings reveal an association between the top transcripts expressed in the biopsy that correlated with dd-cfDNA and IFN-γ and natural killer cells. Additionally, they observed that AKI and fibrosis were associated with mildly elevated dd-cfDNA, and some biopsies had no evidence of rejection despite elevated levels of dd-cfDNA. The researchers suggest the findings indicate plasma dd-cfDNA levels are correlated to the active molecular rejection process in indication biopsies, and dd-cfDNA measurements have the potential to reduce unnecessary biopsies.

The study is well powered and includes a large international cohort of samples that include the simultaneous measurements of both biopsy for transcriptomic profiling and blood for dd-cfDNA measurements. Other strengths of the study include the profiling of sequential biopsies regardless of the results of histology. The findings from this accomplished group of researchers are relevant given the increasing body of literature that suggests molecular measurements need to supplement histologic assessment. The implementation of such a strategy is evident from the recommendations of the Banff Molecular Diagnostics Working Group for the inclusion of the Banff Human Organ Transplant gene expression panel consisting of 770 genes related to rejection, tolerance, and viral infections, and innate and adaptive immune responses.² The hope is that a pathogenesis- and pathway-driven molecular assessment of biopsies will be a useful decision-making tool for treatment, and for the design of clinical trials that require precise phenotyping of biopsies. This is especially important because most diagnostic studies are benchmarked to histologic phenotypes, which, despite their limitations, are still the current “gold standard.”

Although the finding from the study that dd-cfDNA is associated with antibody-mediated rejection (ABMR) is not novel,^3–5 the additional information that higher levels of dd-cfDNA are associated with specific transcripts expressed in the tissue, and its relationship to ABMR, is worthy of further investigation to assess the clinical utility of dd-cfDNA as a biomarker that could potentially replace biopsies, particularly unnecessary ones, which are often the case in a clinical setting. However, the authors must exercise caution in claiming that dd-cfDNA associations with gene expression establishes the clinical utility of dd-cfDNA as a noninvasive marker, especially in the context of indication biopsies where the presence of ABMR and T cell-mediated rejection is already established.^3,6 Similarly, although this group has contributed immensely to the knowledge of biopsy molecular gene expression and its relationship to histologic assessment, the implicit assumption that molecular gene expression is a better gold standard than histology may be overemphasized without adequate validation studies from other groups.

Although the study was predominantly conducted using indication biopsies, the true clinical utility of the relationship between dd-cfDNA and gene expression of correlated transcripts would be from the use of surveillance biopsies. Despite 6% of the study samples being surveillance biopsies (albeit a small number), the lack of a comparison of the performance characteristics between this subset (even if it was a trend) and the indication biopsies is a missed opportunity. The study may also introduce some potential selection bias in that the biopsies ranged from as little as 5 days post-transplant, to as late as 31 years post-transplant. Although this may not have direct bearing on the association between levels of dd-cfDNA with ABMR, the timing of these biopsies may pose a challenge in differentiating what is truly rejection versus injury and fibrosis. This distinction is important in the context of indication biopsies that showed no evidence of rejection, but still had elevated levels of dd-cfDNA. The authors themselves have shown the release of dd-cfDNA by kidneys with no histologic or molecular rejection could reflect parenchymal and fibrotic injury.⁷ If this were true, only an analysis of genes that compares biopsies with high levels of dd-cfDNA (above dd-cfDNA threshold) with biopsies that had rejection and were above the dd-cfDNA threshold can clarify if these differences were due to rejection versus injury. In the absence of such an analysis, one could argue the elevated dd-cfDNA in rejectors may also be due to fibrotic and parenchymal injury, and not active rejection.

It is also not surprising the amounts of ABMR observed in these patient groups were higher than normally seen in a prevalent study, where the early rejection types are predominantly T cell-mediated rejection.^5,8 If the goal is to use dd-cfDNA as a diagnostic tool to detect early rejection, such a test in a scenario such as ABMR, especially late ABMR where there is already substantial damage and fibrosis to the kidney, may not benefit the patient. Similarly, although the study focuses on the relationship between gene expression measurements and dd-cfDNA levels, if the diagnostic is designed to be used to avoid biopsies, then it is essential to provide diagnostic metrics, such as positive and negative predictive values and sensitivity/specificity for the performance of the test in this setting, especially with a clear context of use. This is the minimal amount of information needed to substantiate the claim that the correlation between dd-cfDNA levels and active molecular rejection has the potential to reduce unnecessary biopsies.

In conclusion, although the study provides a valuable contribution to the understanding of the relationship between dd-cfDNA and gene expression in the biopsy, it is limited in its contribution to the understanding of how plasma dd-cfDNA can be used in either detecting rejection or in monitoring patients, and does not substantially change our understanding on the clinical application and the context of use of the dd-cfDNA test.

Disclosures

J.J. Friedewald reports having consultancy agreements with Eurofins Transplant Genomics, Inc., and Sanofi; reports receiving research funding from AbbVie, CSL Behring, Eurofins Viracor, Inc., Hansa BioPharma, Horizon Therapeutics, National Institutes of Health, and Veloxis; reports receiving honoraria from Sanofi; reports having patents or royalties with Northwestern University/Scripps Research Institute; reports having an advisory or leadership role with Eurofins Transplant Genomics, Inc.; and reports receiving speakers bureau from Sanofi. S.M. Kurian reports employment with Scripps Health; reports having consultancy agreements with Transplant Genomics Inc.; reports having patents or royalties with Transplant Genomics Inc.; and reports having an advisory or leadership role with MindX Sciences and Transplant Genomics Inc.

Funding

None.